LO
CapabilityCognitive automation2025-07-29
GPT-4o wrote working G-code for one 3-axis milling part checked in simulation, but complex work still needs CAM programming and expert supervision, a peer-reviewed study found
Machinist / CNC machinistoccupation page →
Event date / reported
2025-07-29
Evidence stage
CapabilityA demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
Tasks this bears on
Programming machines
Turning drawings into toolpaths and machine code (G-code), usually with CAM software, and choosing tools, speeds and feeds.
Being augmented≈ Platform inference
Where this applies
Slovenia, a laboratory study. The authors asked ChatGPT-3.5 and GPT-4o to write ISO G-code for a sample part and checked it in simulation: GPT-4o produced fully functional code, while GPT-3.5's code caused a collision. They conclude that GPT-4o cannot yet replace traditional CAM programming, especially for complex operations, and that it still requires expert supervision. One part, simulated, not cut.
What this means
A general-purpose model can write machine code for a simple part, but people still check it before it reaches a machine.
What it does not yet show
One part checked in simulation; it does not show use in shops or on complex parts.
What you can check
Open the APEM article (published on-line 29 July 2025) and find "still requires expert supervision".
Does it change the assessment?
No — and this stage does not move it either. A "Capability" record is real evidence, but it does not upgrade a task judgement on its own. The 1 linked judgement above stand where they were.
Source
Šket K et al. — G-code generation for 3-axis milling with ChatGPT, Advances in Production Engineering & Management 20(2) (published on-line 29 July 2025) · verified 2026-10-01 · Claude (VOLO agent) · interpreted 2026-10-01 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.